SP1 Initiates Epithelial-Mesenchymal Transition of CTCs and Inhibits Metastasis in Prostate Cancer

Background: Circulating tumor cells (cid:0) CTCs (cid:0) are the basis of cancer metastasis. Till now, the role of different subtypes of CTCs in metastasis is unclear. Methods: We used the Canpatrol TM technique to isolate CTCs from 102 prostate cancer (PCa) patients. The EMT markers of CTCs were detected by FISH and classied CTCs into Epethial (E-CTCs), Mesenchymal (M-CTCs) and Mesenchymal/Epethial-CTCs (M/E-CTCs). Further, the potential EMT related molecules regulators were predicted by bioinformatics software, and SP1 was identied as the key EMT regulator. Then overexpress SP1 of PCa cells to verify the effect of SP1 on the EMT regulation and PCa metastasis. Results: The count of Total-CTCs, E-CTCs, and M/E-CTCs in metastatic PCa was signicantly higher than that in local PCa. Although M-CTCs was not signicantly different between local and metastatic PCa, the ratio of M-CTCs/Total-CTCs in metastatic PCa was markedly lower than that in local PCa. We found that Total-CTCs is an independent risk factor for PCa metastasis. Overexpression of SP1 initiated EMT of PCa cells and enhanced the invasion in vitro. Injecting overexpression SP1 PCa cells via tail vein, generating M-CTCs in vivo, we found the ability of M-CTCs to form lung metastasis was signicantly inhibited compared with that of the control PCa-CTCs. Conclusion: Our study suggested Total-CTCs >14 predict PCa metastasis. M/E-CTCs might facilitate to PCa metastasis; however, M-CTCs might not. SP1 is an EMT regulator, which has the potential role of regulating the EMT of CTCs, thus changing the proportion of subtypes of CTCs. It is a potential therapeutic target. (cid:0) Trial registration: Chinese Clinical Trial Registry. Registered 30 JUNE 2020, http://www.chictr.org.cn/registry.aspx

morphological changes in their phenotype from cuboidal to spindle-shaped, promoting metastasis [10,11]. The speci c role of different CTCs in metastasis remains unclear, which hinders drug development for this mechanism.
Canpatrol TM CTCs isolate technology is a simple, economical, and reliable separation technology based on cell size detecting CTCs in peripheral blood [12]. It is generally believed that cancer cells with mesenchymal phenotype are more aggressive and contribute to metastasis [8,9] . However, their properties would be much different when cancer cells are intravasation to the circulatory system directly. The ability of M-CTCs to cause metastasis may be pretty different in the circulatory system from that in tissues. It is fascinating that our study found the M-CTCs/T-CTCs ratio was signi cantly reduced in metastasis PCa.
Based on the bidirectional reversibility of EMT, CTCs reduce the proportion of M-CTCs with weak metastasis ability, which may be the mechanism of CTCs enhance metastasis.
The expression of intercellular junction molecules ICMs is an essential feature of EMT [13]. When cancer cells start EMT, there are many changes of ICMs. Finding the regulatory genes of these ICMs is of great signi cance for the subsequent development of drugs to control the epithelial or mesenchymal phenotype of CTCs. Therefore, we used bioinformatics methods to analyze the possible ICMs regulatory genes. A total of 111 EMT-related intercellular junction proteins were collected by manual search. Then, we used bioinformatics software to nd out the regulatory molecules of EMT in CTCs. It is deduced that SP1 is one of the critical regulators of these terminal functional molecules. To verify whether this conclusion is effective in CTCs, we detected the expression of SP1 in CTCs of PCa patients and upregulated SP1 in PCa cells. We found that the increase of SP1 level was related to the rise in M-CTCs and M / E-CTCs ratio of T-CTCs. Further experiments con rmed that overexpression of SP1 initiated EMT of PCa cells and inhibited the formation of lung metastases in vivo Fig. 7 . This study con rmed that different varieties of CTCs have different abilities to cause metastasis, and some kinds of CTCs may have a weak ability to drive metastasis. SP1 initiated EMT of PCa cells, which may regulate cancer cells' ability to lead to metastasis. We believe that our results will contribute to developing anticancer drugs targeting CTCs, transforming CTCs into CTCs with weak or no metastasis ability, and providing new ideas for metastatic PCa treatment.

Methods And Materials
Cell culture and Animals Human PCa cell lines PC3 were purchased from ATCC (Manassas, VA). PC3 cells were maintained in DMEM medium (Hyclone, Logan, UT) with 10 % FBS, 100 U/ml penicillin, and streptomycin as the previous study described [14].
Eight-week-old male nude mice (22-26g) necessary for the lung metastasis experiments were purchased from the Animal experiment management centre of Chongqing Medical University (BALB/c background, Chongqing, China). All animal experiments were carried out following guidelines set by Chongqing Medical University (Chongqing, China). We used iso urane inhalation for anaesthesia. Then, Carbon dioxide asphyxiation followed by cervical dislocation was performed for killing.

Patients
The present study is a prospective study conducted from July 2020 to February 2021, including 102 PCa patients. The primary inclusion criteria were histologically con rmed prostate adenocarcinoma diagnosis; no previous androgen deprivation therapy, chemotherapy, surgery, or radiotherapy. Bone scan, Pelvic MRI, chest computerized tomography (CT) or Positron Emission Tomography-Computed Tomography (PET-CT) were used to diagnose metastatic PCa. Our institution's Ethics and Scienti c Committee approved this study, and the enrolled patients provided written informed consent. We registered this study with ClinicalTrials.gov, whose number is ChiCTR2000034274.
Generated SP1 overexpress stable cell lines PC3 cells with stable overexpress of SP1 Lenti-GV-SP1 and respective control cells (control) were generated using GV-492-GFP-Pur lentiviral vector (10 8 p.f.u.) (JiKai, Shanghai China). The stable Lenti-GV-SP1 cell line was developed as the previous study [14]. Brie y, PC3 cells were cultured into 6-well plates and then infected by lent virus solution with polybrene (Sigma-Aldrich) for 16 hours. Three days later, transfection e ciency was tested through GFP expression and subjected to 4 μg/ml puromycin (Life Technologies, Grand Island, NY) selection until GFP was expressed in all cells.

Bioinformatics analysis of regulatory molecules of EMT
We collected about 111 EMT-related molecules by manual search combined with our knowledge background (the list of the genes is available in supplement table1). Online Metascape software (HTTP:// metascape.org/gp/ index.html#/ main/step1 ) was used to analyze the pathway enrichment, Proteinprotein interaction (PPI), and upstream gene regulation of these genes [15,16].

Cancer cells invasion detected by Transwells
Invasion assay was performed using matrigel invasion chambers from BD Biosciences as in a previous study [17]. Brie y, we used 24 wells invasion chamber with 8 um pores coated with Matrigel to detect the invasion. PC3 cells were seeded on the top chambers, then the invaded cells in the lower membrane surface were xed with eosin staining (Olympus, Japan).

Lung metastasis analysis in vivo
We randomly divided nude mice into two groups each group has 5 mice for the administration of PC3 cells. Cells (0.5×10 6 ) suspended in sterile normal saline were injected (i.v.) to each group of mice separately via the tail vein. In all experiments, mice were assessed for metastasis at 2 weeks after PC3 cells were administered. Then lung metastasis was measured by India ink staining as the previous study [14].
After that, the membrane was incubated with secondary antibodies (1: 5000, Abcam, Cambridge, MA). After developing the membrane with ECL reagent (Life Technologies, Grand Island, NY), the protein expression was observed.

Statistical Analysis
All the data are expressed as Mean± SD and were calculated from multiple independent experiments. PSA and Total-CTCs' cutoff value was calculated by Receiver Operating Curve (ROC) curve analysis (Medcalc version 9.6.2.0). The Student's two-tailed t-test was used to determine signi cant differences among groups using SPSS 11.0 software. Regression analysis was used to analyze the independent risk factors of metastatic PCa using SPSS 11.0 software. Data with p< 0.05 were considered to have signi cant differences.

Results
1 The number of CTCs in patients with metastatic prostate cancer was signi cantly higher than that in patients without metastasis.
The separation of CTCs is di cult and expensive, which seriously hinders the research in this eld. Many studies have proved Canpatrol TM CTCs technique to be a safe, economical, and effective method [12].
Although this method is still imperfect, it can fully meet the needs of this study. To verify the method's reliability, we tested three healthy people for CTCs, and the study con rmed that CTCs in healthy people were indeed very low ( Fig. 1A-B).
Although the better way to diagnose metastatic PCa is to use 68 Ga-PSMA-PET/CT, it is costly, and it is challenging to ensure that every patient has the examination. In this study, bone scan, chest CT and Pelvic MRI, or PET-CT were used to determine whether the patient was metastatic PCa. We enrolled 102 PCa patients in the present study. We found that the number of Total-CTCs in patients with metastatic PCa was signi cantly higher than that in patients without metastasis p<0.01, Fig. 1C-D . The number of CTCs in patients with organ metastasis was signi cantly higher than that in patients with bone metastasis p<0.01, Fig. 1E-F . Our study found that the number of E-CTCs and M/E-CTCs in patients with metastatic PCa was signi cantly higher than that in local PCa ( Fig.2C-D, p<0.01). Because of the signi cant difference between the two groups in the total number, we further calculated the proportion of E-CTCs and M/E-CTCs in the Total-CTCs. We found that the ratio of E-CTCs in the total-CTCs was signi cantly lower in patients with metastasis, while the proportion of M/E-CTCs in the total-CTCs in patients with metastasis was signi cantly higher than that in local PCa patients ( Fig.2F-G, p<0.01). There was no signi cant difference in the absolute number of M-CTCs between patients with metastasis and local PCa, but the proportion of M-CTCs/T-CTCs in patients with metastatic PCa was signi cantly lower than that in patients in local PCa (Fig.2B,2E, p<0.01).
3 The number of CTCs is an independent risk factor for prostate cancer metastasis Bone scan, MRI, PET-CT, and even 68 Ga-PSMA-PET/CT are needed to determine the metastasis of PCa.
However, PET-CT and 68 Ga-PSMA-PET/CT are expensive, and patients often do not examine these because of the cost. However, judging whether PCa patients with metastasis have essential value for treatment selecting. Therefore, we intend to explore the predictive effect of CTCs on PCa metastasis We used regression analysis to analyze the predictive effects of age, MRI, Risk strati cation, tumor volume, Gleason score, PSA levels, and Total-CTCs on PCa metastasis. Our study found that PSA and Total-CTCs are independent risk factors for PCa metastasis (Fig.3 A, p<0.01). Through ROC curve analysis, we found that PSA and T-CTCs were good predictors of PCa metastasis. If the count of Total-CTCs > 14, it predicts PCa cells metastasis, in which the sensitivity was 90.48%, and the speci city was 96.67%, more sensitive than PSA (Fig.3 B-C).

SP1 is an essential regulator of cell-cell junction proteins analyzed by bioinformatics software
The change of cell surface molecules is a critical feature of EMT[8]. So far, researchers have reported a large number of EMT surface markers. Complex net mechanisms should regulate these EMT markers. We retrieved 111 reported EMT-related molecules by manual search in Pubmed to nd the main regulatory molecules among numerous cell surface molecules.
We found that these molecules are mainly extracellular matrix organization, cell junction organization, and cell-substrate adhesion molecules by cluster enrichment analysis (Fig.4A). Through the study of protein-protein interaction (PPI), these molecular interactions were divided into three categories according to the degree of interaction, namely, E-Cadherin (CDH1) group molecules, Claudins (CLDNs) group molecules, and FN group molecules (Fig.4B).
To nd these EMT molecules' regulatory molecules, we used TRRUST online software to calculate these molecules' potential upstream regulatory molecules. We discovered that SP1 is the second most involved in EMT regulation, second only to Snail 1. Since many studies reported Snail1 (Snai1) function on EMT already, we choose to explore the role of SP1 in the regulation of EMT of CTCs in PCa. 5 SP1 promotes the transformation of CTCs into mesenchymal CTCs via EMT. SP1 is an essential regulatory molecule of EMT, which can initiate EMT in cancer cells [19]. However, the regulation effect of SP1 on EMT in CTCs is less studied, and the speci c role is still not cleared. In this study, FISH was used to detect the SP1 level of CTCs. In the actual measurement, SP1 was given a purple mark. To show clearly with DAPI, we converted its color to white (Fig.5A).
According to the different expression levels of SP1 in CTCs, we divided them into negative, weakly positive (+), and strongly positive (+ +). We found that in patients with metastatic PCa, the proportion of strongly positive SP1 CTCs increased (Fig.5B). The number of SP1+ cells increased signi cantly in metastatic PCa compared with local PCa; however, there was no signi cant difference in the proportion of SP1 positive cells /Total-CTCs between metastatic PCa and Local PCa. Further data analysis showed that SP1 + + cells/Total-CTCs were signi cantly decreased in metastatic PCa compared with local The results of CTCs only implied that SP1 might be involved in the regulation of EMT in PCa cells. To verify this conclusion, we constructed a PCa cell line, PC3 overexpressing SP1.
Our study con rmed that overexpression of SP1 could reduce the expression of E-Cadherin and increase the expression of N-Cadherin in PCa cell line PC3 (Fig.6 A-D, p<0.05). Overexpression of SP1 in PC3 cells signi cantly enhanced the invasiveness of PC3 cells in vitro (Fig.6 E-F, p<0.05). However, we found that overexpression of SP1 signi cantly inhibited lung metastasis in mice lung metastasis model in vivo (Fig.6 J-H, p<0.05).

Discussion
Circulation Tumor cells (CTCs) are essential for cancer metastasis. A large number of studies have con rmed that EMT of cancer cells is a bidirectional reversible process. We found that among E-CTCs, M-CTCs, and M/E-CTCs, the ability of M-CTCs to induce metastasis is weak. If it promotes CTC's transformation to the type with weak or almost no metastatic ability, it will be a new strategy for treating metastatic cancer. Our study found that overexpression of SP1 initiated the EMT process of CTCs and promoted the transformation of CTCs to M-CTCs, hence inhibiting the metastasis of PCa. The results of this study enrich our knowledge of CTCs and are bene cial for the drug development of metastatic PCa.
Studies have proved that the amount of CTCs is closely related to cancer metastasis, chemotherapy drug resistance, cancer recurrence, and prognosis [20,21]. Our study also con rmed that the number of CTCs in patients with metastatic PCa was signi cantly higher than that in patients without metastasis (Fig.1 A-D). Bone is the most common target organ for PCa metastasis; organ metastasis indicates patients with lung, liver, brain, and kidney metastasis in PCa [22]. The prognosis of prostate cancer patients with organ metastasis is often worse than that of bone metastasis. In patients with metastatic PCa, the quantity of Total-CTCs in patients with organ metastasis is higher than that in patients without organ metastasis ( Fig.1 E-F, p<0.05). These results suggest that the further increase of CTCs may lead to organ metastasis.
The treatment of metastatic PCa is different from that of local PCa. It has essential values to predict whether PCa patients with metastasis. We used regression analysis to analyze the predictive effects of age, MRI, Risk strati cation, tumor volume, Gleason score, PSA levels, and Total-CTCs on PCa metastasis. We found that Total-CTCs is an independent risk factor for PCa metastasis, and its number greater than 14 is highly suggestive of metastasis (Fig.3 A-C). This result may reduce unnecessary PET-CT or 68 Ga-PSMA-PET/CT examination in patients with PCa. However, it needs further study in the future.
EMT is an important mechanism to regulate the metastasis and invasion of cancer cells [23]. The occurrence of epithelial-mesenchymal transition (EMT) in CTCs results in epithelial-(E-CTCs), mesenchymal-( M-CTCs), and hybrid-subtypes of CTCs (M/E-CTCs). So far, the speci c role of different varieties of CTCs in metastasis is still unclear. Our study found that the number of E-CTCs and M/E-CTCs in patients with metastatic PCa was signi cantly higher than that in local PCa (Fig.2C-D, p<0.01), suggesting these two kinds of CTCs might be facilitated to PCa metastasis. Unlike in vitro experiments, it is generally believed that the migration and motility of epithelial cancer cells will be weakened [24]. With the deepening of research, it has also been reported that the overexpression of epithelial marker molecules in cancer cells is necessary for cell metastasis, which implies that Epithelial-CTCs may contribute to metastasis [25].
Many studies reported that cancer cells transformed into mesenchymal cells would obtain more vigorous invasion and have a more substantial metastasis potential in vitro. However, in the circulatory system, the role of M-CTCs may be different from that of in vitro experiments. Moreover, as known to us all, cells' invasiveness is different from patients' metastatic ability. In animal experiments, few studies suggested that M-CTCs may have a weak ability to cause metastasis [26]. This study found no signi cant difference in the absolute number of M-CTCs between patients with metastasis and local PCa. However, if further comparing the ratio of M-CTCs, we found that the proportion of M-CTCs in Total-CTCs of patients with metastatic PCa was signi cantly lower than those in local PCa (Fig.2B,2E, p<0.01). The main reason why there was no statistical difference in the absolute number of M-CTCs between the local and metastatic PCa was that the absolute number of M-CTCs was too low to distinguish statistically. Combined with the results of animal experiments in this study, the lung metastasis caused by M-CTCs was signi cantly lower than that caused by E-CTCs. Based on these results, we speculate that there may be some mechanism to reduce the proportion of M-CTCs in T-CTCs in patients with metastatic PCa, and increase the absolute number of E-CTCs and M/E-CTCs on the premise of increasing T-CTCs, to promote the PCa metastasis.
In this study, bioinformatics technology was used to screen potential EMT regulatory molecules of CTCs. Then we double-checked to verify whether they are EMT-related molecules by cluster enrichment analysis. The results showed these molecules are mainly extracellular matrix organization, cell junction organization, and cell-substrate adhesion molecules (Fig.4A). Through the analysis of PPI, we found that these molecular interactions can be divided into three categories according to the degree of interaction, namely, E-Cadherin (CDH1) group molecules, Claudins (CLDNs) group molecules, and FN group molecules (Fig.4B). The above analyses con rmed that the genes we used were EMT-related once again. Finally, we selected SP1 molecules by TRRUST analysis, which is less studied. SP1 is a zinc nger transcription factor that binds to GC-rich motifs of many promoters. The encoded protein is involved in many cellular processes, including cell differentiation, cell growth, apoptosis, immune responses, response to DNA damage, and chromatin remodelling. Some studies demonstrate SP1 activates EMT in cancer cells, but few studies on the function of EMT regulated by SP1 in CTCs, especially in PCa. This study found a good correlation between the number of SP1 positive cells and M-CTCs+M/E-CTCs count by correlation analysis (Fig.5 G-J). Along with the increase of SP1 + and SP1++, CTCs tended to transform to M/E-CTCs or M-CTCs+ M/E-CTCs Fig.5 F . These results suggest that SP1 might be involved in the EMT regulation of PCa CTCs, and the increase of SP1 level may promote the transformation of CTCs cells to mesenchymal phenotype.
To verify the effect of SP1 in PCa patients, we overexpressed SP1 in PC3 cells. We found that overexpression of SP1 reduced the expression of E-Cadherin and increased the expression of N-Cadherin in PC3 (Fig.6 A-D, p<0.05), which con rmed that overexpression of SP1 initiated EMT of PCa cells and promoted the mesenchymal transformation of cancer cells. Functional studies have also demonstrated that the overexpression of SP1 in PC3 cells signi cantly enhanced the invasiveness of PC3 cells in vitro (Fig.6 E-F, p<0.05). However, we found that overexpression of SP1 signi cantly inhibited lung metastasis in mice lung metastasis model in vivo (Fig.6 J-H, p<0.05). Interestingly, the results of in vivo and in vitro experiments did not match; that is, M-CTCs are not conducive to the formation of cancer metastasis but enhance the invasion of cancer cells. It has been reported that overexpression of twist, one of EMTtranscription factors (EMT-TFs), maintains the cancer cells in mesenchymal phenotype and signi cantly inhibits cancer cells' lung metastasis formation through tail vein injection [26,27]. These results indicated M-CTCs are weak to cause metastasis, and up-regulation of SP1 in CTCs could inhibit metastasis.

Conclusion
In conclusion, the count of CTCs is an independent prognostic factor for PCa metastasis. Reducing the proportion of M-CTCs and increasing the count of E-CTCs and M/E-CTCs might promote metastasis. SP1 initiates EMT to inhibit CTCs' metastasis ability. SP1 may become the therapeutic target for PCa patients. Indeed, there are still some de ciencies in this study. The speci c roles of E-CTCs and M / E-CTCs in metastasis and related regulatory mechanisms need to be further studied.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.    The number of CTCs is an independent risk factor for prostate cancer metastasis. A, Regression analysis con rmed that PSA and Total-CTCs were independent risk factors for PCa metastasis. B-C, The cut-off value of PSA and Total-CTCs to predict metastatic PCa by ROC analysis.